Diametros nr 7 (marzec 2006): 37 – 55
Conceivability, Possibility, and A Posteriori
Necessity: on Chalmers’s Argument for Dualism
Karol Polcyn
1.
Introduction
Chalmers argues that zombies are possible and that therefore consciousness does
not supervene on physical facts, which shows the falsity of materialism1. The
crucial step in this argument – that zombies are possible – follows from their
conceivability and hence depends on assuming that conceivability implies
possibility. But while Chalmers’s defense of this assumption – call it the
conceivability principle – is the key part of his argument, it has not been well
understood. As I see it, Chalmers’s defense of the conceivability principle comes in
his response to the so-called objection from a posteriori necessity. The defense
aims at showing that there is no gap between conceivability and possibility since
no such gap can be generated by necessary a posteriori truths2. I will argue that
while Chalmers is right to the extent that there is no gap between conceivability
and possibility within the standard Kripkean model of a posteriori necessity, his
general conclusion is not justified. This is because the conceivability principle
might be inconsistent with a posteriori necessity understood in some nonKripkean way and Chalmers has not shown that no such alternative
understanding of a posteriori necessity is available.
1
See Chalmers [1996].
Anthony Brueckner is one of the authors that acknowledge the key importance of Chalmers’s
discussion of a posteriori necessity in Chalmers’s defense of the conceivability principle. See
Brueckner [2001]. Brueckner argues that Chalmers’s defense begs the question, but in my view
Brueckner’s diagnosis is not correct. As I will argue in what follows, although there is questionbegging in the part of the argument that Brueckner refers to, the key argument is different.
2
37
Karol Polcyn
2.
Conceivability, Possibility, and A Posteriori Necessity...
Two-dimensional semantics
As Chalmers observes, the objection from a posteriori necessity is the most natural
objection to the conceivability principle. The point of the objection is that the
conceivability principle is inconsistent with the phenomenon of a posteriori
necessity discovered by Kripke. Referring to the Kripkean a posteriori necessities,
such as “Water is H2O”, one might think that we have clear examples of truths
that are conceivable but a posteriori or ‘metaphysically’ impossible, such as
“Water is not H2O”3.
Chalmers argues that the objection from a posteriori necessity is not a good
objection against the conceivability principle. Since he discusses this objection
within the framework of two-dimensional semantics, let me begin with laying
down some key distinctions that are made within that framework. According to
the two-dimensional semantics, there are two intensions associated with each
concept: primary intension and secondary intension, where intension is
understood as a function from possible worlds to reference. The primary intension
picks out the reference of a given concept in a possible world considered as actual,
whereas the secondary intension picks out reference in a possible world
considered as counterfactual. The important point is that these two intensions turn
out to be different in the case of natural kind concepts, such as ‘water’. Whereas
according to its primary intension, ‘water’ refers to the kind of stuff that is watery
(has the manifest properties of water) in a given possible world considered as
actual (to H2O in our world and XYZ, say, in another possible world), the value of
its secondary intension remains the same across all possible (counterfactual)
worlds and depends on what water actually is. Thus, assuming that water is H2O
in the actual world, ‘water’ refers, according to its secondary intension, to H2O in
every counterfactual world. Hence, water is necessarily H2O or, in other words, it
is impossible for water not to be H2O according to the secondary intension of
‘water’, whereas it is possible that water is not H2O if we think of water in terms of
the primary intension of ‘water’.
3
See Kripke [1980].
38
Karol Polcyn
Conceivability, Possibility, and A Posteriori Necessity...
The reason why the primary intension of ‘water’ is different from its
secondary intension has to do with the fact that ‘water’ is a rigidified definite
description. ‘Water’ picks out its reference in the actual world under the
description “the dominant clear, drinkable liquid in the oceans and lakes (more
briefly: ‘watery stuff’)” and that description might have actually been satisfied by
a kind of stuff that is different from what actually fits the description4. However,
given that the description associated with ‘water’ is rigidified, ‘water’ will pick out
in counterfactual possible worlds only one kind of stuff, namely one that satisfies
the description in the actual world5.
Further, Chalmers points out that in the case of natural kind concepts the
primary intension is determined a priori:
The primary intension of a concept, unlike the secondary intension, is independent
of empirical factors: the intension specifies how reference depends on the way the
external world turns out, so it does not itself depend on the way the external world
turns out6.
The secondary intension, by contrast, does depend on how the actual world turns
out: it depends on what actually satisfies the description associated with a given
natural kind concept. So the secondary intension of natural kind concepts,
according to Chalmers, is determined a posteriori.
It is worth emphasizing that Chalmers’s distinction between the a priori
and the a posteriori needs to be taken in this context very carefully. In particular,
the claim that primary intension is determined a priori is not as obvious as
Chalmers thinks. It is true that which values the primary intension of ‘water’, say,
takes in various possible worlds does not depend on the a posteriori fact that
watery stuff is H2O. From this, however, it does not follow that the primary
intension of ‘water’ is determined a priori, without making use of any a posteriori
4
See Chalmers [1996] p. 57.
By contrast with natural kind concepts, theoretical concepts, such as ‘H2O’ or ‘XYZ’, are rigid
without being rigidified descriptions. Therefore, in their case the primary and secondary intensions
coincide.
5
6
Chalmers [1996] p. 57.
39
Karol Polcyn
Conceivability, Possibility, and A Posteriori Necessity...
considerations. As Block and Stalnaker argue, there are limits as to what sort of
watery stuff could count as water in the actual world and those limits are not
determined a priori but depend on our theories and methodological principles, in
fact, on everything we believe7. We now know that watery stuff is H2O and we
think of that stuff as identical with water. If we were to find out that the colorless,
odorless, drinkable stuff in rivers and lakes is XYZ, we would conclude that water
is XYZ. If we found out that such stuff is not really a liquid, we would have found
out that water is not a liquid. Certainly, though, we can also imagine finding about
watery stuff something totally unexpectable to normal English speakers,
something that might force us to reject the supposition that watery stuff is water.
Thus, not everything that is a priori possible will correspond to what is possible
according to the primary intension of the word ‘water’ as we use it.
Chalmers’s assumption to the effect that primary intension is determined a
priori is not only false but would lead to question-begging in the defense of the
conceivability principle. For to assume that primary intension is determined a
priori is to assume that what is primarily possible follows from what is a priori
conceivable and that is precisely what is at issue. Now, Anthony Brueckner thinks
that Chalmers does beg the question in defending the conceivability principle8. For
according to Brueckner, the key move that Chalmers makes in based precisely on
assuming that primary intension is determined a priori. That is, according to
Brueckner, Chalmers’s reasoning is as simple as that: zombies are possible since
they are possible according to the primary intension of the concept of
consciousness and that they are primarily possible follows from the fact that they
are a priori conceivable.
I do not think that Brueckner’s reconstruction of Chalmers’s reasoning is
fair, at least it is not a good reconstruction of Chalmers’s response to the objection
from a posteriori necessity. If we think that Chalmers takes it for granted that a
priori conceivability implies possibility according to primary intension, we must
see Chalmers as defending the conceivability principle directly. But that is not, in
7
See Block and Stalnaker [1999].
8
See Brueckner [2001].
40
Karol Polcyn
Conceivability, Possibility, and A Posteriori Necessity...
fact, what Chalmers is doing. Instead, Chalmers defends his conceivability
principle indirectly, by arguing that the objection from a posteriori necessity is not
a good objection against that principle. The interesting question then is whether
Chalmers has shown that. I will argue that he has, at least if we understand a
posteriori necessity in the standard Kripkean way.
Secondly, it needs to be pointed out that it would not be justified at this
point to argue against Chalmers that the conceivability principle must be wrong if,
contrary to what Chalmers believes, primary intension across possible worlds is
not determined a priori. One might think that if primary intension is determined a
posteriori, in Block’s and Stalnaker’s sense, what is conceivable might nevertheless
be impossible according to that intension. But that does not follow. The a
posteriori constraints over what is possible according to primary intension are
constraints on the proper use of concepts and not on what is possible per se. It is
certainly conceivable, to apply Block’s and Stalnaker’s example to the case of
watery stuff, that watery stuff might have actually been an extinct flightless bird9.
But even though we would not say, in the light of our knowledge and beliefs, that
this kind of watery stuff might have been water, as Block and Stalnaker suggest,
this does not mean that the conceived situation is impossible but only that it
would not be correct to describe it as true of water. As I will argue, Chalmers gives
a similar sort of response to the objection from a posteriori necessity – the sort of
necessity that arises at the level of secondary rather than primary intension.
3.
Chalmers’s response to the objection from a posteriori necessity
Equipped with his two-dimensional semantics, Chalmers argues that the objection
from a posteriori necessity mentioned earlier is not a good objection against the
conceivability principle. Without denying that possibility (in a broad metaphysical
and not only nomological sense) depends on facts that are accessible a posteriori,
Chalmers argues that those facts constrain only the way in which conceivable
worlds should be described, without affecting their possibility.
Block and Stalnaker use the example of an extinct flightless bird as the possible reference of the
English word ‘coumarone’. Coumarone, we are told, is another colorless liquid, not H2O but
C6H4OCHCH. See Block and Stalnaker [1999].
9
41
Karol Polcyn
Conceivability, Possibility, and A Posteriori Necessity...
In order to understand Chalmers’s response to the objection from a
posteriori necessity, we need to get clear as to what the objection really amounts to
within the two-dimensional semantics. What does it mean to say, in terms of that
semantics, that something might be conceivable and yet a posteriori impossible,
for example, that it is conceivable but not possible that water is not H2O but XYZ,
say? While the impossibility of water being XYZ is cashed out within that
semantics as the necessary falsehood of the statement “Water is XYZ” when
evaluated according to its secondary intension, its being conceivable that water is
XYZ amounts simply to the idea that the statement “Water is XYZ” is conceivably
true (as opposed to possibly true) according to the primary intension of ‘water’10.
That statement could not be conceivably true according to the secondary intension
of ‘water’ for the secondary intension of ‘water’ picks out H2O in all conceivable
worlds and it is inconceivable that H2O be XYZ. However, it is conceivable that
watery stuff might have been XYZ and that is what the conceivability of the
statement “Water is XYZ” amounts to when we evaluate that statement according
to the primary intensions of the notions involved. Thus, coming back to the
objection from a posteriori necessity, the point of the objection must simply be
this: what is conceivable according to primary intension might be impossible
according to secondary intension.
Chalmers‘s response to this objection is made most explicit when Chalmers
says that “nothing about Kripke’s a posteriori necessity renders any logically
possible worlds impossible”11. I reconstruct this response as follows. The fact that
the statement “Water is XYZ” is necessarily false when evaluated according to its
secondary intension does not make impossible any of the conceivable worlds in
which the statement “Water is XYZ” is true when evaluated according to its
primary intension. The conceivable worlds, in which this statement is true,
according to the primary intensions of the concepts involved, are the worlds in
Conceivable worlds are as good arguments of intensions as possible worlds and Chalmers
himself speaks of the distinction between primary and secondary intensions also in the context of
conceivable worlds, although he does not explicitly express the objection from a posteriori
necessity in the way I am suggesting. See Chalmers [1996] p. 65-69.
10
11
Chalmers [1996] p. 134.
42
Karol Polcyn
Conceivability, Possibility, and A Posteriori Necessity...
which watery stuff is XYZ. But since the secondary intension of ‘water’ is ‘H2O’
and not ‘watery stuff’, the fact that “Water is XYZ” is necessarily false according to
the secondary intension of ‘water’ does not mean that it is impossible for watery
stuff to be XYZ but only that it is impossible for H2O to be XYZ. Thus, instead of
making any conceivable worlds in which watery stuff is XYZ impossible, the
impossibility of H2O being XYZ only implies that it would not be correct to
describe those worlds as ones in which water, according to the secondary
intension of ‘water’, is XYZ. In this sense, as Chalmers says, the a posteriori
necessity does not put a posteriori constraints on the space of possible worlds but
merely constrains the way in which certain terms are used to describe it.
So Chalmers’s response to the claim that it is conceivable but not possible
that water is XYZ amounts to the simple idea that what is impossible in the case of
water is not quite the same as what is conceivable. To say that it is conceivable that
water is XYZ is to say that watery stuff might have been XYZ, whereas to say that
it is impossible that water be XYZ is to say that it is impossible for H2O to be XYZ.
Since the subject matter of these two claims is different, they do not conflict with
each other and hence do not provide the counterexample to the conceivability
principle. The whole distinction between its being conceivable that water is XYZ
and its being impossible that water be XYZ is only a distinction at the level of
statements, as Chalmers puts it, and not at the level of worlds.
Now, having established that the necessary identity of water and H2O does
not undermine the conceivability principle, Chalmers assumes that no example of
a posteriori necessity that we are familiar with can do that. This is because all
known cases of a posteriori necessity can be explained within the two-dimensional
framework in the same way as the case of water and H2O, that is, by assuming
that there are two different intensions associated with the relevant natural kind
concepts. Hence, in all other cases the distinction between conceivability and
possibility will be only a distinction at the level of intensions or statements.
To summarize, Chalmers’s response to the objection from a posteriori
necessity is as follows:
43
Karol Polcyn
Conceivability, Possibility, and A Posteriori Necessity...
(1) The Kripkean a posteriori necessity attaches to a posteriori identity statements
and only if the secondary intension of at least one of the concepts flanking the
identity sign is different from the primary intension12.
(2) What is impossible according to secondary intension is not conceivable, so it
cannot be said that the Kripkean a posteriori necessity makes impossible what is
conceivable according to secondary intension. On the other hand, the Kripkean a
posteriori necessity does not make impossible what is conceivable according to
primary intension, either. For if the secondary and primary intensions of the
concept F are different, the distinction between its being conceivable that F is not
G and its being impossible that F is not G is only a distinction at the level of
statements.
(3) So the Kripkean a posteriori necessity does not make any conceivable truths
impossible.
There is no question-begging in Chalmers’s reasoning as reconstructed above.
Without assuming that conceivability implies possibility, Chalmers has shown
that the objection from a posteriori necessity does not work as an objection against
the conceivability principle.
This result, according to Chalmers, defends the possibility of zombies
against the objection that zombies might be a posteriori or metaphysically
impossible despite their conceptual coherence. Since in all cases of a posteriori
necessary identities that we know of the space of possible worlds is not a
posteriori constrained, then, by analogy, the possibility of zombies should not be a
posteriori constrained by the a posteriori and necessary identity, if any, between
consciousness and some physical property, either. The point is that even if such
identity were true, it could only make a semantical and not a metaphysical
difference, that is, it could only constrain the way in which zombies should be
described, without affecting their possibility.
A posteriori necessity results within the Kripkean framework from rigidifying at least one of the
concepts flanking the identity sign and, as I pointed out earlier, rigidification is responsible for the
divergence of the primary and secondary intensions of the concept rigidified.
12
44
Karol Polcyn
4.
Conceivability, Possibility, and A Posteriori Necessity...
Chalmers’s response elaborated further: strong metaphysical necessity
To be sure, however, Chalmers has only shown at this point that the conceivability
principle is not undermined by the Kripkean model of a posteriori necessity. But
why should we think that this is the only model of a posteriori necessity available?
If there is an alternative model, one that puts a posteriori constraints on the space
of possible worlds, the conceivability principle would be false.
The alternative account should allow the possibility that identity statements
be necessary and a posteriori also in the case when the primary and secondary
intensions of both concepts flanking the identity sign are identical. Within the
Kripkean framework, an a posteriori identity statement can be necessary only in
the case when the primary and secondary intensions of at least one of the concepts
flanking the identity sign are different and, as Chalmers has shown, due to this
difference of intensions the relevant necessity does not generate any conflict
between conceivability and possibility at the level of worlds so that no impossible
truth of the form (F is not G) is conceivable or, in other words, no conceivable
truth of the form (F is not G) is impossible. But in the alternative model, the
relevant conceivable truths of the form (F is not G) would be impossible. To make
it clear, assume that the identity statement “F = G” is a posteriori and necessary so
that it is impossible that F is not G and that the primary and secondary intensions
of both ‘F’ and ‘G’ coincide. In this case, assuming that it is conceivable that F is
not G, we could not interpret the distinction between its being conceivable that F is
not G and its being impossible that F is not G as a distinction at the level of
statements or intensions. We would have to say that the very situation that is
conceivable is impossible and not just that it is not correct to describe it in a certain
way. So under the assumption that the primary and secondary intensions of both
‘F’ and ‘G’ coincide, the relevant a posteriori necessity would put a posteriori
constraints on the space of possible worlds and the conceivability principle would
be false.
Of course, the alternative model of a posteriori necessity would apply to the
case of statements of psychophysical identity since the primary and secondary
intensions of phenomenal and physical concepts coincide. Thus, even though we
45
Karol Polcyn
Conceivability, Possibility, and A Posteriori Necessity...
find psychophysical identity statements conceivably false according to their
primary intensions, those statements will have to be necessary when interpreted
according to their primary intensions after all if we assume that they are necessary
according to their secondary intensions. This affects, of course, the possibility of
zombies. As we saw, Chalmers argues that, assuming that the primary and
secondary intensions of given concepts differ, what is necessarily true according to
secondary intension cannot affect what is possible according to primary intension.
Thus, he says that even if we assume that zombies are impossible according to the
secondary intension of the concept of consciousness, they may still be possible
according to the primary intension of the concept of consciousness. Of course,
Chalmers can say that only assuming that the primary and secondary intensions
of the concept of consciousness differ. If we assume that the primary and
secondary intensions of the concept of consciousness are the same, zombies will
not be primarily possible after all13.
Now, Chalmers is well aware of the objection that we could account for the
necessity of psychophysical identity within a non-Kripkean model of a posteriori
necessity. In response to this objection, then, Chalmers simply argues that the
alternative model, construed in the way described above, is, indeed, impossible.
The alternative model would generate what Chalmers calls ‘strong metaphysical
necessity’ as opposed to the Kripkean ‘weak’ necessity that only constrains the
way in which possible worlds should be described. Chalmers then argues that the
idea of strong necessity is unacceptable. In The Conscious Mind Chalmers gives the
following reasons against that idea:
13 Chalmers seems to be confused about it in The Conscious Mind. On page 132 he says that his
argument for dualism will go through if we concentrate on the primary intension of the concept of
consciousness and ignore the secondary intension. Then, on page 133, he says that “the irrelevance
of a posteriori necessity can be further supported by the observation that with consciousness, the
primary and secondary intensions coincide.” Chalmers’s dialectic here seems to be the following.
Zombies are possible according to the primary intension of the concept of consciousness since they
are conceivable according to that intension. And once it has been granted that zombies are
primarily possible, they then must be seen as possible according to the secondary intension of the
concept of consciousness, too, assuming that the primary and secondary intensions of that concept
are identical. The first move, of course, begs the question against materialists who deny the
conceivability principle.
46
Karol Polcyn
Conceivability, Possibility, and A Posteriori Necessity...
(1) Strong necessity will put constraints on the space of possible worlds that are
brute and inexplicable. While there may be brute and inexplicable facts about our
world, the realm of the possible has no room for this sort of arbitrary constraint;
(2) The idea of strong necessity cannot be supported by analogies since all
necessary truths we are familiar with are the Kripkean necessities of the sort
“Water is H2O” or “Hesperus is Phosphorus” and those necessities require only a
single space of worlds;
(3) There cannot be a posteriori constraints on the space of possible worlds since a
posteriori information can only tell us about our world14;
(4) What is logically possible (conceivable) is conceptually coherent and
presumably it is in God’s power to do anything that is coherent, hence it is
presumably in God’s power to create a zombie world. However, the advocate of
strong necessity must say either that the possibility is coherent but God could not
have created it, which is unjustified, or that God could have created it but it is
nevertheless metaphysically impossible, which is entirely arbitrary;
(5) Even if you believe in strong necessity, it remains the case that facts about
consciousness cannot be derived from physical facts. So in order to explain the
supervenience of the phenomenal on the physical you need to introduce certain
primitive connecting principles. But then for all explanatory purposes you are left
in the same position in which property dualism leaves you;
This point has been emphasized all along in the recent discussion of a posteriori necessity. One
might think that the existence of necessary a posteriori truths implies that there is some sort of
necessity that is accessible only a posteriori. Thus, one might think that the necessity of the
statement “Water = H2O”, say, must be accessible a posteriori since the sentence “Water = H2O” is
a posteriori true. But that is a mistake. The a posteriori status of “Water = H2O” is a status of that
sentence in the actual world, whereas necessity is a matter of truth across all possible worlds. Thus,
from the fact that “Water = H2O” is a posteriori true it certainly does not follow that the necessity
of that statement is accessible a posteriori, too. In fact, we cannot infer that water is necessarily H2O
merely from the fact that “Water = H2O” is a posteriori true. We need to assume in addition that
‘water’ is a rigid designator, that it designates in all possible worlds only one kind of stuff, namely
one that is watery in the actual world. But once that assumption is in place, it then becomes a
conceptual truth that if H2O is the actual watery stuff, then water is H2O across all possible worlds.
That is to say, the inference is licensed by the way we use the concept of water. See Jackson [1997].
14
47
Karol Polcyn
Conceivability, Possibility, and A Posteriori Necessity...
(6) The sort of materialism that is based on the idea of strong necessity is far more
mysterious than the dualist alternative. The invocation of brute principles
constraining the space of possible worlds introduces an element much more
problematic, and indeed far les naturalistic, than the mere invocation of further
natural laws postulated by property dualism.
Now, I agree with Chalmers that the idea of strong necessity is far from obvious.
However, I do not think that any of Chalmers’s objections mentioned above
undermine the intelligibility of the idea that psychophysical identity statements
can be necessary and a posteriori on the assumption that the primary and
secondary intensions of the concept of consciousness are identical. Think about
such statements independently of Chalmers’s objections. If there is any difficulty
in viewing such statements as a posteriori and necessary, it seems that it cannot lie
in thinking that they should be necessary if they are true a posteriori. That
inference seems to be justified by the fact that both phenomenal and physical
concepts are rigid. So assuming that statements of psychophysical identity can be
true a posteriori, they simply can be a posteriori and necessary, and I do not see
why any of Chalmers’s objections mentioned above should rule that out.
But let’s look at the objections systematically. First, the necessity of
psychophysical identity within the alternative model of a posteriori necessity does
not seem to be brute and inexplicable, as Chalmers suggests. The alternative
model provides an explanation of how the necessity arises. According to this
model, the necessity of psychophysical identity simply follows from assuming that
the identity is a posteriori true and assuming that the primary and secondary
intensions of phenomenal and physical concepts, respectively, are the same. There
seems to be nothing mysterious or inexplicable about that inference. Secondly, the
mere fact that the necessity of psychophysical identity cannot be supported by
standard a posteriori necessary identities of the Kripkean sort does not carry any
weight. We have perfectly good reasons to think that the necessity of
psychophysical identity is sui-generis since the semantics of phenomenal concepts
is radically different form the semantics of other natural kind terms; unlike in the
case of other natural kind concepts, the primary and secondary intensions of
48
Karol Polcyn
Conceivability, Possibility, and A Posteriori Necessity...
phenomenal concepts coincide15. Thirdly, we are not assuming that the a
posteriori identities (psychophysical identities) that constrain the space of possible
worlds are true of other possible worlds a posteriori; that they are true of other
possible worlds follows a priori from the fact that the primary and secondary
intensions of phenomenal and physical concept, respectively, are the same. As for
the forth objection, it needs to be clarified what Chalmers means by conceptual
coherence in this context. If he means to say that if something is conceivable, it is
rational for us to believe it is possible, then we need not assume that what is
conceivable is always conceptually coherent. For again, within the alternative
model of a posteriori necessity it is not rational after all to think that conceivability
implies possibility. As for the fifth objection, we are not assuming that
phenomenal facts supervene upon physical facts; rather we are taking
phenomenal facts to be identical with physical facts. And finally, as for the sixth
objection, the alternative model of a posteriori necessity does not look mysterious;
this is the point that was already made in response to the first objection.
Chalmers emphasizes that the fundamental problem with the idea of strong
necessity is that this sort of necessity breaks the natural tie between rationality and
modality. As Chalmers says, there is a circle of interrelated modal notions, such as
possibility, consistency and rational entailment, and the framework of possible
worlds is supposed to make sense of those notions. But the introduction of strong
necessity does not help in making sense of those notions, according to Chalmers.
This is precisely because the tie between rationality and modality is broken within
the framework of strong necessity. The tie is broken since strong necessity breaks
the tie between conceivability and possibility.
Clearly, Chalmers assumes in this line of reasoning that there is a tie
between conceivability and rationality so that if something is conceivable it is
rational or conceptually coherent to believe it is possible. But then what he sees as
the fundamental objection against the idea of strong necessity really amounts to
the forth objection discussed above and, as we saw, the force of this objection is
questionable.
15
This has been emphasized by Brian Loar. See Loar [1997].
49
Karol Polcyn
5.
Conceivability, Possibility, and A Posteriori Necessity...
The a posteriori status of psychophysical identity
As I suggested, once we assume that psychophysical identity is true a posteriori, it
is hard to see why it should fail to be true necessarily according to both the
secondary and primary intensions of the notions involved. However, one might
argue that the problem with psychophysical identity statements is that they
cannot be true a posteriori. In fact, this is what Chalmers suggests in his later
formulations of the argument from the conceivability of zombies. Thus, already in
Mind and Modality Chalmers assumes explicitly that an identity statement can be
true a posteriori only if it has a contingent primary intension and that claim, if
true, would rule out the possibility that psychophysical identity statements be true
a posteriori16. However, Chalmers’s defense of his assumption is not convincing.
Chalmers defends his assumption indirectly, by arguing that its denial would be
unacceptable since it would commit us to believing in strong metaphysical
necessity and strong necessity is itself unacceptable, as Chalmers argues, for the
reasons that he already gave in The Conscious Mind. But clearly, whether or not
strong metaphysical necessity is unacceptable is precisely what is at issue. And as
we just saw, the reasons that Chalmers gives in The Conscious Mind are not good
reasons against believing in strong necessity.
Given that there seems to be nothing wrong with assuming that certain
identity statements, in particular psychophysical identity statements, can be
necessary in the strong sense if we allow them to be true a posteriori, the defense
of the claim that such statements cannot be true a posteriori cannot depend on
assuming that they cannot be necessary in the strong sense. So here is the key
question: why should we think that psychophysical identity statements cannot be
true a posteriori? The potential difficulty seems to be this. Given that the primary
and secondary intensions of both phenomenal and physical concepts, respectively,
coincide, the properties that fix the reference of those concepts are identical with
the referents of those concepts. Thus, if we assume that phenomenal and physical
concepts pick out the same properties, we would have to assume that the
16
See Chalmers [1998].
50
Karol Polcyn
Conceivability, Possibility, and A Posteriori Necessity...
properties that fix the reference of phenomenal concepts are identical with the
properties that fix the reference of physical concepts. In this respect, psychophysical identity statements would differ from standard a posteriori identities,
such as “Water = H2O” or “Heat = molecular motion”. For the properties that fix
the reference of ‘water’ (watery properties) are different from the property that
fixes the reference of ‘H2O’ (the property of being composed of H2O molecules)
and the property that fixes the reference of ‘heat’ (heat sensations) is different
from the property that fixes the reference of ‘molecular motion’ (the property of
being molecular motion). Now, why should this difference between standard a
posteriori identities and psychophysical identity carry any weight against the
latter? You might think as follows. The a posteriori truth of standard theoretical
identities is justified by the discovery of certain a posteriori relations between the
properties that fix the reference of the relevant concepts. Thus, the identity of
water and H2O is true since the fact that water is composed of H2O molecules
explains why water is watery, and the identity of heat and molecular motion is true
since molecular motion causes heat sensations. But there is no room for the
corresponding a posteriori relations between the properties that fix the reference
of theoretical-physical concepts and the properties that fix the reference of
phenomenal concepts since, by assumption, those properties are identical.
I do not see why this difficulty should carry much weight. Granting that
there is no a posteriori relation of the relevant sort between the properties that fix
the reference of phenomenal and physical concepts, we can justify the a posteriori
truth of psychophysical identity by appealing to the fact that phenomenal and
physical states always appear at the same place in the causal scheme of things17.
Phenomenal states have certain behavioral effects and we are justified in assuming
that those are the effects of certain physical states since we can explain the
occurrence of those effects in physical terms. Thus, we are justified in assuming on
a posteriori grounds that phenomenal states and the relevant physical states are
identical.
17
This point is emphasized by David Papineau. See Papineau [1999].
51
Karol Polcyn
Conceivability, Possibility, and A Posteriori Necessity...
There seems to be a residual difficulty. Given that the properties that fix the
reference of phenomenal and physical concepts are identical, we might expect that
we should be able to see a priori, simply in virtue of understanding those
concepts, that they pick out the same properties. However, many philosophers, in
particular Brian Loar, argue that this expectation – the expectation of transparency, as
Loar calls it – is an illusion. Phenomenal and physical concepts have radically
different cognitive roles and this difference keeps them unconnected a priori
despite the fact that the properties fixing the reference of those concepts are
identical18.
It is worth pointing out that the above response drawing on the nature of
phenomenal concepts does not by itself justify the claim that phenomenal and
physical concepts pick out the same properties. This is, in fact, what Chalmers
complains about in his response to Loar. That phenomenal and physical concepts
have different cognitive roles shows that those concepts cannot be connected a
priori but that does not yet guarantee that those concepts have the same
referents19. Chalmers’s point is, of course, right. However, there is nothing to complain about here. Loar’s point about the different cognitive roles of phenomenal
and physical concepts is not meant to be the complete account of the coreference
of those concepts. The point is only meant to explain away the expectation that we
should be able to see a priori that phenomenal and physical concepts have the
same referents if they do. The explanation of why those concepts actually corefer
has to be supplemented by some independent reasons. As I mentioned above, it
seems natural to suggest that the explanation in question can be based on the fact
that phenomenal and physical states play the same roles in the causal scheme of
things.
Let me review the key points of the last two sections. We have been
wondering whether identity statements whose primary and secondary intensions
coincide, in particular statements of psychophysical identity, can be necessary and
a posteriori. If they can, the relevant identities will put a posteriori constraints on
18
See Loar [1997].
19
See Chalmers [1999].
52
Karol Polcyn
Conceivability, Possibility, and A Posteriori Necessity...
the space of possible worlds. Of course, full discussion of whether the sorts of
statements in question can be necessary and a posteriori goes beyond the scope of
this paper. But whether or not those statements can be necessary and a posteriori,
it seems clear that Chalmers himself leaves us with no reason to think that they
cannot. This is important in our context for this means that Chalmers is not
justified in assuming that there can be no a posteriori constraints on the space of
possible worlds. Thus, it remains an open question whether or not there are a
posteriori constraints on the possibility of zombies. If psychophysical identity
statements can be necessary and a posteriori, zombies will be impossible.
6.
Conclusion
Chalmers argues that the conceivability principle is not inconsistent with the
phenomenon of a posteriori necessity. I have argued that while Chalmers is right
in so far as a posteriori necessity is understood in the standard Kripkean way, his
argument is not conclusive because it is far from obvious that the Kripkean model
of a posteriori necessity is the only model of a posteriori necessity we can think of.
To assume that it is, as Chalmers does, is to assume that an a posteriori identity
statement can be necessary only in the case when the primary and secondary
intensions of at least one of the concepts flanking the identity sign are different.
My point is that Chalmers has not explained why that should be true. Thus, it
remains possible that certain identity statements, in particular psychophysical
identity statements, can be true a posteriori and necessary even though the
primary and secondary intensions of both concepts flanking the identity sign in
those statements coincide. If so, the conceivability principle would be inconsistent
with such truths.
Clearly, what is really at stake in Chalmers’s defense of the conceivability
principle against the objection from a posteriori necessity is whether or not
psychophysical identity can be true a posteriori and necessary. If Chalmers is right
about his assumption that a posteriori necessity requires the divergence of
primary and secondary intensions, then statements of psychophysical identity
cannot be both true a posteriori and necessary. Thus, if Chalmers’s assumption is
53
Karol Polcyn
Conceivability, Possibility, and A Posteriori Necessity...
true, it would create a problem for physicalism regardless of whether the
conceivability principle as such is true for then it would follow that the alleged
necessary relation between consciousness and the physical world could be
justified neither a posteriori nor a priori; the lack of a priori justification follows
from the mere conceivability of zombies.
Paradoxically, then, Chalmers’s argument for dualism does not require the
conceivability principle as such. The way Chalmers initially formulates the
challenge for physicalism comes down to the question as to whether or not
zombies are possible. If the question is put this way, Chalmers has to defend the
conceivability principle since the only evidence for the possibility of zombies he
can offer is their conceivability. However, while defending the conceivability
principle against the objection from a posteriori necessity Chalmers assumes
implicitly that the Kripkean model of a posteriori necessity is the only model we
can think of, and since that model is not applicable to psychophysical identity
statements, Chalmers’s assumption leads directly to the antiphysicalist conclusion
to the effect that there cannot be the relation of a posteriori and necessary identity
between consciousness and the physical world. As I have argued, however,
Chalmers’s assumption in question is left without sufficient justification and for
this reason Chalmers has not shown that psychophysical identity cannot be
necessary and a posteriori.
References
Block and Stalnaker [1999] – N. Block and R. Stalnaker, Conceptual Analysis, Dualism, and
the Explanatory Gap, “The Philosophical Review” (108) 1999.
Brueckner [2001] – A. Brueckner, Chalmers’s conceivability argument for dualism, “Analysis”
(61) 2001.
Chalmers [1996] – D. Chalmers, The Conscious Mind, Oxford University Press, Oxford
1996.
Chalmers [1998] – D. Chalmers, Mind and Modality, http://consc.net./chalmers/
Chalmers [1999] – D. Chalmers, Materialism and the Metaphysics of Modality, “Philosophy
and Phenomenological Research” (59) 1999.
54
Karol Polcyn
Conceivability, Possibility, and A Posteriori Necessity...
Jackson [1997] – F. Jackson, Finding the Mind in the Natural World, [in:] The Nature of
Consciousness, ed. N. Block, The MIT Press, Cambridge, Mass. 1997.
Kripke [1980] – S. Kripke, Naming and Necessity, Harvard University Press, Cambridge,
Mass. 1980.
Loar [1997] – B. Loar, Phenomenal States, [in:] The Nature of Consciousness, ed. N. Block, The
MIT Press, Cambridge, Mass. 1997.
Papineau [1999] – D. Papineau, Mind the Gap, “Philosophical Perspectives” (12) 1999.
55